• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 52 subscribers
  • Views 203 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Issues comparing Count Event "breakpoint" against System Analyzer

Nick Borton
Genius 3595 points

Hello,

I'm currently working on optimizing a high frequency interrupt on a C6472. I'm attempting to understand why it is taking so long. System analyzer states that it takes on average 4us to complete. When I turn on a hardware count event which counts cpu cycles between the beginning and end of my ISR (inclusive of subroutines) it comes back with 470 cpu cycles. With a 625MHz clock, that is 5x faster than 4us.

Is the count even acting like I expect? based on system performance, the 4us from system analyzer looks correct.


I was hoping to avoid placing lots of Log_write statements in my code, and just use the same binary and move the count event start and stop locations.

Thanks

  • 0
    TI__Guru**** 171034 points

    Hi,

    What are the releases of CCS, the TI emulators package and the manufacturer emulator package (Spectrum Digital, Blackhawk or other)? You can find this info under the menu Help --> About --> Installation Details button.

    Also, how are you setting up the hardware count event? Is it the "Profile Cycles" event via the Breakpoints view or using the Profiler Clock? Do you have a test case that you could send?

    Unfortunately I don't have a C6472 board with me and I am having trouble finding some some code that has both System Analyzer and a hardware interrupt for my C6455 device (the closest to C6472 I have).

    I will keep trying to find a suitable example code or something that could have been reported.

    Regards,

    Rafael

  • 0 in reply to desouza
    Genius 3595 points

    CCS: 5.3

    TI Emulators: 5.0.872.0

    Spectrum Digital: 5.2.0.03

    Here is the breakpoint type I am using:

    My guess is that you don't have to use a HW interrupt to verify it. a SWI with a bunch of function calls nested 2 to 3 deep would do it.

    Thanks